Nucleoside analogs are an important class of compounds that have potential utility as treatments for a variety of diseases. Some nucleoside analogs may be efficacious because of their ability to inhibit adenosine kinase. Adenosine kinase catalyzes the phosphorylation of adenosine to adenosine 5'-monophosphate (AMP). Inhibition of adenosine kinase may effectively increase the extracellular level of adenosine in humans and thereby potentially serve as a treatment of ischemic conditions such as stroke, inflammation, arthritis, seizures, and epilepsy. See e.g. U.S. Pat. No. 5,674,998. Nucleoside analogs may also have potential for the treatment of chronic pain. However, in order to conduct clinical trials to determine the clinical efficacy of nucleoside analogs as well as meet the demand once the therapy is brought to the market, scaled-up quantities of the purified nucleoside analog of interest are necessary.
The nucleoside analogs that can be synthesized by the process of this invention consist of a furanose covalently bound to a heterocycle (B) as represented by the following formula: ##STR1##
The substitutents at the C4' position of the furanose ring, R.sub.1 and R.sub.2, can be optionally and independently substituted by groups such as (C.sub.1 -C.sub.6) alkyl or substituted (C.sub.1 -C.sub.6) alkyl. If R.sub.1 is --CH.sub.2 --OH and R.sub.2 is --H, then the nucleoside sugar is ribose. The furanose contains an anomeric carbon (at position C1'). Because of the anomeric carbon, there are two stereoisomers, or anomers, of furanose: the .alpha.-anomer and the .beta.-anomer. See e.g. Streitweiser and Heathcock, 1985, Introduction to Organic Chemistry, Macmillan. The stereoisomerism of furanose results in corresponding .alpha. and .beta. nucleoside isomers. Typically, the .beta.-nucleoside anomer is the anomer of biological interest.
The synthesis of pure .beta.-nucleosides and .beta.-nucleoside analogs has proven to be difficult. Many published schemes for nucleoside analog synthesis result in .alpha.:.beta. anomeric mixtures. Typically, it is difficult to isolate the .beta.-anomer from such .alpha.:.beta. mixtures, especially if economical scaled-up quantities of pure .beta.-nucleoside analog are required. To avoid undesirable .alpha.:.beta. nucleoside analog mixtures, the art has emphasized the use of starting reagents that favor the production of .beta.-nucleoside analogs over .alpha.-nucleoside analogs. The prior art can be divided into three categories (i) heavy metal approaches (ii) .alpha.-furanosyl halide approaches, and (iii) a sodium hydride/N,N-dimethylformamide based approach.
The heavy metal approaches use a heteroanion coupled with a heavy metal such as mercury or silver. For example, Lerner was able to use a heavy metal approach to produce yields of 9-(.alpha.,.beta.)-L-erythrofuranoslyadenine in which the .beta.-anomer predominated over the .alpha.-anomer by a ratio of almost thirty to one. Lerner, 1969, Carbohydr. Res., 9, 1-4. In the Lerner method, a chloromercuriheterocycle complex was reacted with a .beta.-furanosyl chloride in the presence of a hot hydrocarbon solvent to yield the nucleoside analog. This reaction was driven by the formation of mercuric dichloride salt. Because of the toxicity of mercury, the Lerner approach is not optimal for large-scale synthesis of .beta.-nucleosides. However, it is well known in the art that silver can often be used instead of mercury in reactions such as those provided by Lerner et al. Thus the toxicity of mercury can be avoided.
In general, the heavy metal approaches are generally adequate but less than satisfactory in practice. Stoichiometric quantities of heavy metal, relative to the sugar or the heterocycle, are required. In scaled-up reactions, the use of 1:1 molar ratios of heavy metal to heterocycle anion is expensive and typically does not provide an economical solution for the stereospecific synthesis of biologically active nucleoside analogs. Furthermore, in the case of some nucleoside analogs of interest, the heavy metal approaches often fail altogether.
The .alpha.-furanosyl halide approaches use .alpha.-furanosyl halide as a starting reagent. In such approaches, an inversion at the anomeric center of the .alpha.-furanosyl chloride occurs upon coupling of the furanosyl with a heteroanion. For example Ramasay et al. stereospecifically synthesized .beta.-7-deazaguanosine and related nucleoside analogs using an .alpha.-furanosyl halide approach. Ramasay et al., 1987, Tetrahedron Letters, 28, 5107-5110. The method of Ramasay et al. is illustrated in scheme (I): ##STR2##
where TBDS represents t-butyldimethylsilyl. In reaction scheme (I), the sodium salt of heterocycle (2) was generated by treatment with sodium hydride (NaH) in acetonitrile and then reacted with .alpha.-furanosyl chloride (1) to afford protected .beta.-nucleoside analog (3). The protected .beta.-nucleoside analog (3) was then deprotected to yield .beta.-nucleoside (4). The .alpha.-furanosyl chloride (1) may be prepared according to the methods of Wilcox et al., 1986, Tetrahedron Lett., 27, 1011.
Like the heavy metal approaches, the .alpha.-furanosyl halide methods are functional but not always satisfactory in practice. For instance, such methods are not always amendable to scale-up because the .alpha.-furanosyl halide starting material is typically unstable and rapidly isomerizes to the .beta.-isomer. Furthermore, the .alpha.-furanosyl halide is more difficult to form than the .beta.-furanosyl halide anomer. For instance, special reaction conditions, such as low temperatures, are typically needed to synthesize the .alpha.-furanosyl halide. For these reasons, approaches that use a .beta.-furanosyl halide anomer as a starting reagent are preferred over .alpha.-furanosyl halide approaches.
The sodium hydride/N,N-dimethylformamide approach does not involve the use of heavy metals or .alpha.-furanosyl halides. Kondo et al., 1986, Tetrahedron, 42, 199-205. In Kondo, an .alpha.:.beta. mixture of nucleoside analog (8) was synthesized by coupling .beta.-furanosyl chloride (5) with the heterocycle anion of (6) to yield the protected .beta.-nucleoside analog (7) according to scheme (II): ##STR3##
wherein Tr represents trityl and DMF represents N,N-dimethylformamide. The protected .beta.-nucleoside analog (7) was then deprotected using standard methods to yield the .beta.-nucleoside analog (8). Kondo et al. found that the coupling of the .beta.-furanosyl chloride (5) with heterocycle (6) in the presence of sodium hydride and DMF resulted in a highly undesirable 3:1 mixture of (.alpha.,.beta.)-nucleoside analog (8). To improve the .alpha.:.beta. anomeric selectivity, Kondo et al. experimented with the use of NaBr, NaI, MgBr.sub.2 OEt.sub.2, or (n-Bu).sub.4 NBr as additives to reaction (II). The best results that Kondo et al. obtained based upon this experimentation was a 1:2 .alpha.:.beta. mixture of nucleoside analog (8) when the coupling reaction (II) was carried out in the presence of sodium hydride, powdered sodium bromide, and DMF.
As demonstrated by Kondo et aL., the NaH/DMF approach is generally undesirable for scaled-up production of .beta.-nucleoside analogs because the method does not selectively provide .beta. nucleoside analog product. Rather, a mixture of the .alpha. and .beta. anomers of the nucleoside analog product is formed. Thus, .beta. nucleosides synthesized by the NaH/DMF method must be purified from an .alpha.:.beta. nucleoside analog mixture. Such a purification step may be particularly difficult if large scale synthesis of nucleoside is desired. In summary, the NaH/DMF method of Kondo et al. does not adequately address the need in the art for an economical method for selectively synthesizing .beta.-nucleosides and .beta.-nucleoside analogs.
As described above, the cited references refer to numerous methods for making .beta.-nucleoside analogs. Each method, while functional in many situations, has drawbacks. Some nucleoside analogs cannot be synthesized by the prior art methods. Often the prior art methods do not provide very good .alpha.:.beta. anomeric selectivity. Many of the prior art methods require starting materials that require special synthetic steps and/or are unstable. Accordingly, there is a need in the art for an improved, scalable method for selectively synthesizing the .beta. anomer of nucleosides and nucleoside analogs.